Novel synthetic procedures for C2 substituted imidazoquinolines as ligands for the α/β-interface of the GABAA-receptor

A series of substituted imidazoquinolines, a structurally related chemotype to pyrazoloquinolinones, a well-known class of GABAA ligands, was prepared via two synthetic procedures and the efficiency of these procedures were compared. One method relies on classical heterocyclic synthesis, the other one aims at late-stage decoration of a truncated scaffold via direct C–H functionalization. A pharmacological evaluation disclosed that one of the synthesized derivatives showed interesting activity on a α1β3 containing receptor subtype. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00706-022-02988-8.


Introduction
The endogenous neurotransmitter, γ-aminobutyric acid (GABA), plays a major role in neurotransmission in the mammalian brain, where it binds to ligand-gated ion channel type A receptors (GABA A ) [1,2].These pentameric receptors are composed of different subunits (α1-6, β1-3, γ1-3, δ, ε, π, θ, ρ1-3) and the specific assemblies thereof (receptor subtype) determine their pharmacological effects [3].Upon GABA binding, an anion flux is induced, which can be affected by a variety of allosteric modulators.This is of clinical importance in various ways.For example, it has been reported that sedative and anxiolytic effects are elicited by allosteric modulators (such as benzodiazepines), depending on the particular receptor subtype [4].
Allosteric modulators which use a large number of allosteric sites [5] are in widespread use as anaesthetics, anticonvulsants, sedatives, and tranquilizers [6,7].Moreover, they are known to bind to the α + /γ-interfaces of the pentameric receptors [8].Nevertheless, the treatment with these drugs is accompanied by several severe side effects, such as addictive behaviour and withdrawal syndromes [9].Hence, there is an urgent need for new pharmaceutical ligands with a reduced side effect profile.
Another class of exogenous ligands for the GABA A receptors are pyrazoloquinolinones which showed interesting modulatory activities via a different binding site, namely the α + /β-interface [10].However, these compounds still possess a promiscuous binding profile [2] and poor solubility in most polar solvents.Thus, they are investigated to 1 3 become pharmacological tool compounds by variation of their substitution pattern [11][12][13][14][15][16].
In line with the concept of pharmacophore modelling, the class of imidazoquinolines are a closely related chemotype to the pyrazoloquinolinones and differs mainly in the change of one hydrogen bond acceptor entity (Fig. 1).Therefore, they represent a promising class of compounds in terms of GABA A receptor activity [17].Since synthetic routes towards this peculiar tricyclic scaffold are scarce [18][19][20], we wanted to investigate an alternative method to gain flexibility in synthesis.High functional group tolerance was the main prerequisite for a newly developed method.Moreover, we aimed to investigate the modulatory activity of the newly synthesized compounds at a GABA A receptor subtype.

Results and discussion
To allow the introduction of various substituents on ring D, a retrosynthetic analysis of lead structure I suggested two main building blocks: the diamino quinoline II and benzaldehyde or benzoic acid III (Scheme 1).This route allows late-stage modifications of substituents R 1 on ring D via the use of different benzaldehyde or benzoic acid derivatives.However, the diamino quinoline II has to be synthesized for each substituent R 2 .
The alternative retrosynthetic cut b (Scheme 1) leads to IV and an appropriated substituted benzene V. Following this pathway, a direct arylation in position 2 can be envisioned taking advantage of transition metal catalysed C-H functionalization.In addition, in this case the corresponding building block IV would need to be prepared for each substituent R 2 separately.Hence the difference lies in the final introduction of the aryl-moitey at C2. Additionally, it has to be kept in mind, that the C-H coupling at C2 might be in competition with N-arylation.To determine, which method would be the overall more efficient one, both approaches were investigated.
For both approaches, diaminoquinolines have to be prepared as a crucial intermediate.Synthetic routes towards these compounds have been published previously and were also established partially in our group [12].
The aniline starting materials 1a-1d determine the substitution pattern on ring A in the final molecule.Building up on our findings from the study on pyrazoloquinolinones we decided to use the unsubstituted aniline 1a, two halogenated anilines 1b and 1c and anisidine 1d to introduce a methoxy substituent [12].The first step of the synthesis was the condensation of the anilines 1a-1d to the quinolones 3a-3d.This was done in a two-step process.In the first step, the anilines 1a-1d reacted with diethylmalonate in a Gould-Jacobs reaction [21] to form the corresponding enamines 2a-2d in excellent yields (90%-quantitative).Then, after a solvent exchange to a higher boiling solvent, the ring closure to quinolones 3a-3d was performed.In this step, high temperatures were required (> 150 °C) since intermediately aromaticity is lost.Not surprisingly, such high temperatures led to the formation of a significant amount of (unidentified) side products.Nevertheless, the desired carboxylated quinolones 3a-3d were obtained in acceptable yields (31-61%) and good purity, as they were hardly soluble in any organic solvent and could be easily purified by trituration.The decarboxylation of the ester group was achieved also in a twostep process: first, the ester was hydrolysed and after the carboxylic acid was isolated by precipitation, it was thermally decarboxylated to yield the quinolones 4a-4d, again by using Ph 2 O as a high-boiling solvent (Scheme 2).
Then, we introduced the nitro group in position 3 which was accomplished by standard nitration conditions using a mixture of nitric and acetic acid which gave the 3-nitroquinolones 5a-5d in good yields.Subsequently, chlorination was performed using POCl 3 , yielding the chlorinated nitroquinolines 6a-6d (Scheme 3).
The final steps towards the diamino quinolines now differed for halogen-bearing derivatives 6b and 6c, as milder reduction conditions had to be employed to avoid dehalogenation.Therefore, the amino group on position 4 was introduced directly using aq.ammonia in 1,4-dioxane, which gave the amino-nitro quinolines 7a-7c in quantitative yields.Reduction of the nitro group was carried out under modified Béchamp conditions [22] using iron and NH 4 Cl to give diamino quinolines 8a-8c (Scheme 4).If other reduction conditions, like palladium on charcoal under H 2 atmosphere were employed, substantial dehalogenation was observed.
In cases where dehalogenation was not an issue, a route via azide 7d can be applied.Subsequent reduction using palladium on charcoal under a hydrogen atmosphere afforded the desired diamino quinoline 8d (Scheme 5).
The diamino quinolines 8a-8d now enabled the syntheses of the desired imidazoquinolines.Initially, compounds 9a-9d were prepared to test the direct arylation approach.Cyclization of 8a-8d with formic acid led to the target compounds in good to excellent yields (Scheme 6, 57-98%).
Optimization of the direct arylation was carried out using 9a as substrate, as the simplest derivative of this series.First, we employed a protocol established by Bellina et al. for imidazoles, azoles, indoles, and benzimidazoles [23].There, the C-H activation is carried out by a palladium-copper complex under ligand-and base-free conditions in DMF.When we applied the same conditions on 9a, the direct arylation towards the desired product 10a occurred, but only in 15% yield after a reaction time of 48 h (Table 1, entries 1 and 2).In this first experiment, we observed two main issues which decreased the formation of product.Most importantly, N-arylation of the nitrogen in position three indeed took place and led to significant side product formation.Furthermore, the solubility of the starting material 9a was low and precipitation occurred during the reaction.
Therefore, the catalytic system by displacement of palladium with a different transition metal species (Rh and Ni catalysts, see SI for details).In addition, the influence of Scheme 2

Scheme 3
different bases as well as ligands and additives was investigated (Table 1, entries 3 and 4, and SI).As this only had limited success, we built upon our initial successful transformation using Pd(OAc) 2 and CuI in DMF and focused on finding conditions which prevent the precipitation of starting material.This might be caused by the formation of a copper salt, which has been reported previously for similar heterocyclic systems [24].Therefore, we tried to reduce the amount of the copper source to a minimum which still allows the transformation.We found that a decreased amount of copper additive (0.6 equiv.)under prolonged reaction times (96 h) led to an increased isolated yield of 34% of the desired 10a (Table 1, entry 6).Further attempts to increase the yield by N-protection to suppress side product formation and to increase solubility failed (see SI).
Consequently, we switched to a more classical cyclization approach using either carboxylic acids or aldehydes as reaction partners.Benzaldehyde was used in combination with nitrophenol (as solvent and oxidant) to obtain the desired imidazoquinolines 10a and 10c [25].The same conditions using now p-formylbenzonitrile could be applied to introduce a cyano group on ring D in the final product.This afforded the desired products 11a, 11b, and 11d in good yields (Scheme 6).However, the same methodology led only to the intermediary formed imine with p-methoxybenzaldehyde.Thus, we decided to use the more stable carboxylic acid under harsher conditions.Polyphosphoric acid, a strong acidic and hygroscopic reagent, which additionally binds the formed water allowed the synthesis of the methoxy substituted imidazoquinolines 12b and 12d.
Overall, we synthesized a library of 11 imidazoquinolines which we aimed to investigate their activity as allosteric modulators in GABA A receptors (Fig. 2).
The modulatory activity was investigated by testing the ligands in recombinant α1β3 GABA A receptors.The receptors were expressed in X. laevis oocytes and the compounds were screened at 1 µM and 10 µM at 3-5% (EC 3-5 ) of the maximum GABA elicited current (see methods).
Compounds 11a, 11b, 11d (cyano substituted D ring) and 12b, 12d (methoxy substituted D ring) were tested to compare their activity with similarly substituted pyrazoloquinolinones [7,8].Interestingly, compound 11a displayed a significant modulatory activity (Fig. 3), while the other compounds were not sufficiently active to further characterize in this receptor subtype.Thus, compound 11a represents to the best of our knowledge the first imidazoquinolone which significantly modulates GABA A receptors independently of the presence of a benzodiazepine binding site.

Conclusion
We were able to establish a versatile synthetic route towards the class of imidazoquinolines, which already showed interesting activity on the tested GABA A receptor subtype.Compound 11a gave the most promising results and is the first example of a positive allosteric modulator of its class.The synthetic route was adapted in a way to allow the introduction of a large number of substituents, especially via the first shown CH activation of an imidazoquinoline towards 3-arylimidazoquinolines.Additionally, it is important to note that the imidazoquinoline products 10, 11, and 12 displayed increased solubility as compared to the corresponding pyrazoloquinolinones. Future research will be directed towards further improving the activity profile and establish the imidazoquinolones as subtype-selective GABA A receptor ligands.

Experimental
All starting materials and reagents were purchased from commercial sources and used without further purification.Reactions were monitored by TLC on silica gel 60 F254 plates.Normal-phase column chromatography was performed on silica gel 60 (230-400 mesh).NMR spectra were recorded at 297 K in the solvent indicated, with 200, 400, and 600 MHz instruments, respectively, employing standard software provided by the manufacturer. 1H NMR and 13 C NMR spectra were referenced to tetramethylsilane (TMS, δ = 0) by calibration with the residual organic solvent signals [26].Accurate mass analysis (2 ppm mass accuracy) was carried out from 10 to 100 mg/dm 3 solutions via LC-TOFMS measurements using an autosampler, an HPLC system with binary pumps, degasser, and column thermostat and ESI-TOF mass spectrometer.Melting points were determined with a Büchi Melting Point B-545 apparatus with a heating rate of 1 °C/min (70% onset point and 10% clear point) or on a Kofler Block apparatus.All melting points were obtained without additional recrystallization directly after flash column chromatography (FCC) with light petroleum (LP) and EtOAc and subsequent drying in a high vacuum.

Two-electrode voltage clamp electrophysiology
All steps were performed as reported previously [14].cDNA vectors were linearized, transcribed and purified to generate mRNAS, which were used for injecting Xenopus laevis oocytes.For the microinjection, the RNA of the α1β3 receptor combination was mixed at 1:1 ratio with a final concentration of 56 ng/mm 3 .Oocytes were obtained from commercial or local academic suppliers.Stage 5-6 oocytes with the follicle cell layer around them were roughly dissected with forceps, digested with collagenase (type IA, Sigma, NO, 1 mg/ml ND96 [96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES; pH 7.5)] at 18 °C shaking at 70 rpm for 20-40 min and gently defolliculated with the aid of a pipette and a platinum loop.For electrophysiological recordings, oocytes were placed on a nylon-grid in a bath of Ca 2+ -containing NDE solution medium [96 mM NaCl, 5 mM HEPES-NaOH (pH 7.5), 2 mM KCl, 1 mM MgCl 2 ‧6H 2 O, 1.8 mM CaCl 2 ‧2H 2 O].For current measurements, the oocytes were impaled with two microelectrodes (1-3 MΩ) filled with 2 M KCl.The oocytes were constantly washed by a flow of 6 cm 3 /min NDE that could be switched to NDE containing GABA and/or drugs.Drugs were diluted into NDE from DMSO-solutions resulting in a final concentration of 0.1% DMSO perfusing the oocytes.DMSO was used as blind control.Compounds were co-applied with GABA until a peak response was observed.All recordings were performed at room temperature at a holding potential of − 60 mV using a Dagan TEV-200A two-electrode voltage clamp (Dagan Corporation, Mineapolis, MN).Data were digitized, recorded and measured using an Axon Digidata 1550 low-noise data acquisition system (Axon Instruments, Union City, CA).Data acquisition was done using pCLAMP v.10.5 (Molecular Devices™, Sunnyvale, CA).Data were analysed using GraphPad Prism v.6.and plotted as bar diagrams/bar graphs.Data are given as mean ± SEM from at least three oocytes of two and more oocyte batches.

General procedure a: synthesis of malonates
According to a modified literature procedure [27], anilines 1a-1d (1.00 equiv.)and diethyl-2-ethoxymethylene malonate (1.00 equiv.)were dissolved in toluene (1.25 cm 3 / mmol).The reaction was heated to reflux, up to 22 h, until full conversion was observed by TLC (LP/EtOAc, 3:1).The reaction mixture was cooled to rt and the solvent was removed under reduced pressure.The oil residue obtained was lyophilized to give the desired products 2a-2d in adequate purity.

General procedure B: cyclization to quinolones
According to a modified literature procedure [28], malonates 2a-2d (1.00 equiv.)were dissolved in diphenyl ether (2 cm 3 / mmol), the atmosphere changed to argon and the reaction mixture was heated to reflux for one hour.The reaction time did not allow full conversion but disabled the formation of side products.The reaction mixture was cooled to rt and poured into LP to precipitate the desired quinolone which was collected by filtration and washed several times with a mixture of LP/EtOAc (1:1) to obtain the desired products 3a-3d.

General procedure C: decarboxylation
According to a modified literature procedure [29], quinolone carboxylates 3a-3d (1.00 equiv.)were suspended in 2 N NaOH solution (20 cm 3 /mmol).The reaction mixture was heated to reflux, up to 4 h, until full conversion was observed by TLC (CH 2 Cl 2 /MeOH, 9:1).The reaction mixture was cooled to rt and neutralized with 2 N HCl to precipitate the product.The product was collected by filtration, washed with 200 cm 3 water and dried.Decarboxylation was performed in Ph 2 O (20 cm 3 /mmol).Carboxylic acid (1.00 equiv.) was suspended and heated to reflux (250 °C) up to 2 h until full conversion was observed by TLC (CH 2 Cl 2 /MeOH, 9:1).The reaction mixture was cooled to rt and poured into LP to precipitate the desired products 4a-4d which were collected by filtration, washed several times with LP and dried.

General procedure D: nitration
According to a modified literature protocol [30], quinolones 4a-4d (1.00 equiv.)were dissolved in AcOH (15 cm 3 /mmol) under heating.Concentrated HNO 3 (2.2 equiv.) was diluted with AcOH (1:10) and added to the reaction mixture dropwise.The reaction mixture was heated to reflux for up to 4 h until the consumption of starting material was observed by TLC (CH 2 Cl 2 /MeOH, 9:1).The reaction mixture became orange colored.After cooling to rt, the reaction was poured into water, whereupon the product precipitated.The precipitate was collected by filtration and washed with small amounts of EtOH and water and dried to yield the desired products 5a-5d.

General procedure E: chlorination of nitroquinolones
According to a modified literature protocol [31], nitroquinolones 5a-5d (1.00 equiv.)were dispersed in POCl 3 (4.00equiv.).The reaction mixture was heated to reflux for up to 4 h, until full consumption of the starting material was observed by TLC (LP/EtOAc, 3:1).The reaction mixture was poured on a small amount of ice whereupon precipitation occurred and neutralized with sat.aq.NaHCO 3 .The aqueous layer was extracted with CH 2 Cl 2 , washed with brine, dried over Na 2 SO 4 and the solvent evaporated.The products 6a-6d were purified by FCC (gradient of 5%-15% EtOAc in LP).

General procedure F: synthesis of nitroquinolin-4-amines
According to a modified literature procedure [32], nitroquinolones 6a-6c (1.00 equiv.)were dissolved in 1,4-dioxane (10 cm 3 /mmol) and aq.NH 4 OH (25%, 10 cm 3 /mmol) was added.The reaction mixture was heated to reflux up to 1 h, until full consumption of starting material was observed by TLC (CH 2 Cl 2 /MeOH, 19:1).After cooling to rt, a precipitate was formed.The solvent was evaporated and the precipitate was dissolved in EtOAc and extracted with water and brine, dried over Na 2 SO 4 and evaporated to give the desired products 7a-7c.
The reaction was then filtered through a bed of Celite.

Fig. 1
Fig.1The change of the hydrogen bond donor entity is highlighted in blue and indicated by the green arrow Scheme 4